Re: [cmis-browser] Ordery Schema and referencing

[Gregory Melahn <melahn@us.ibm.com>]

Thanks Jens,

I don't have any ideas except just denormalizing and not using refs at all - and that would be very cumbersome,

I did reach out to Lloyd Hilaiel again this morning to ask him to look at my previous email which was just to confirm he was ok with inclusion of Orderly in the appendix. From the first response I got from him in which he said "I would be willing to work with you to make sure you could use orderly for this project in the way you want with a minimal amount of legal clutter." it seems he is flexible, but I am trying to close on the details. This is what I had said in the earlier emaill (the one he has not replied to yet).

Let's give him a day to respond to my reminder, and if that does not work, I would propose to use orderly@librelist.com to communicate ideas and questions

Hello Lloyd,

Thank you for the quick answer!

First, just to clarify, I don't propose to use any of the code though of course any of the vendors that implement or use the spec may be interested. But that is not really a spec issue.

As far as the design, what I would like us to do is just copy the description of Orderly into our specification, so for example we would have an Appendix A including the text from http://orderly-json.org/docs copied into it. Then, elsewhere in the specification we would use Orderly as a way of describing the details of the binding and we would even include an Orderly file downloadable from the OASIS site containing our schema. I am still having a conversation with the OASIS general counsel about this concept and maybe he will find some flaw in this idea but so far he thinks it is doable assuming you are in agreement of course. And of course we would want to include in our specification the right attribution of Orderly The reason for including the text in the spec itself and not just referring to http://orderly-json.org/docs is to make the specification self contained and not dependent on another document that is itself not (yet) a standard. Would you find such a way of using Orderly acceptable? Or do you have another idea?

Also, in the course of creating our schema we would want to bounce questions off you, or ideas for extending Orderly. For example, one of the committee members thought it would be useful to have a construct of map, and had a proposed syntax for that. Would you be agreeable to members of the subcommittee contacting you with such questions or ideas? We want to respect your time and how you would like such a conversation to work (using a mailing list for example?).

Finally, I would like to ask you to review our proposed schema from time to time and give us ideas for improvement.

To give you an idea on our timeline, the browser binding extension to CMIS is part of the CMIS 1.1 work and the expectation is that we would turn the proposal over to the editors by roughly March 2011 and then it goes into the OASIS spec editing and review cycle which will take most of 2011. So we would want our work on the description of Orderly and the schema complete by March 2011.

So how does all this sound?

Gregory Melahn
STSM
IBM Collaboration Solutions
melahn@us.ibm.com
919 254 0295


Jens Hübel ---11/30/2010 10:45:54 AM---Hi all,

From:	Jens Hübel <jhuebel@opentext.com>
To:	<cmis-browser@lists.oasis-open.org>
Date:	11/30/2010 10:45 AM
Subject:	[cmis-browser] Ordery Schema and referencing

---

Hi all,

I am still not sure how well the referencing in Orderly is currently defined. As Orderly translates to JSON-Schema I looked at this spec. Not the latest draft, but this one here has helpful examples: http://groups.google.com/group/json-schema/web/json-schema-proposal---second-draft (see Extending and Referencing). 

The idea seems to be to have an id either containing a name or a URL and use this in a ref-tag.

Translated to Orderly and the repositoryInfo as an example this would look like:

object {     
		 id "http://docs.oasis-open.org/ns/cmis/browser/201103/repositoryCapabilities";                                
      string  capabilityACL ["none", "discover", "manage"];
      boolean capabilityAllVersionsSearchable;
		 ...
 }*;

object {
      id "http://docs.oasis-open.org/ns/cmis/browser/201103/repositoryInfo"
		 string repositoryId;
      string repositoryName;
      string repositoryDescription;
		 ...
      ref "http://docs.oasis-open.org/ns/cmis/browser/201103/repositoryCapabilities" capabilities;
		 ...
 }*;

The problem is that the "id" currently is not part of Orderly (why?). Orderly allows extensions that are directly passed to JSON-Schema:

object {
		 string myext `{"id": " http://docs.oasis-open.org/ns/cmis/browser/201103/repositoryCapabilities"}`;     
      string  capabilityACL ["none", "discover", "manage"];
      boolean capabilityAllVersionsSearchable;
		 ...
 }*;


But this is only allowed on properties level not on top-level. If we have to copy+paste the Orderly spec anyway I wonder if it makes sense to extend it by this id mechanism. Perhaps we should first ask the Orderly experts for an explanation why they did not do this. 

Has anyone further ideas in this area?

It would be cool to have the complete browser binding schema in a parseable file usable by validation tools.

Jens


---------------------------------------------------------------------
To unsubscribe from this mail list, you must leave the OASIS TC that
generates this mail.  Follow this link to all your TCs in OASIS at:
https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php